|
| 1 | +use harper_core::{CharStringExt, Mask, Masker, Span}; |
| 2 | + |
| 3 | +/// Masker for selecting portions of Literate Haskell documents. |
| 4 | +/// |
| 5 | +/// Based on the specifications outlined at https://wiki.haskell.org/Literate_programming. |
| 6 | +pub struct LiterateHaskellMasker { |
| 7 | + text: bool, |
| 8 | + code: bool, |
| 9 | +} |
| 10 | + |
| 11 | +impl LiterateHaskellMasker { |
| 12 | + pub fn text_only() -> Self { |
| 13 | + Self { |
| 14 | + text: true, |
| 15 | + code: false, |
| 16 | + } |
| 17 | + } |
| 18 | + |
| 19 | + pub fn code_only() -> Self { |
| 20 | + Self { |
| 21 | + text: false, |
| 22 | + code: true, |
| 23 | + } |
| 24 | + } |
| 25 | +} |
| 26 | + |
| 27 | +impl Masker for LiterateHaskellMasker { |
| 28 | + fn create_mask(&self, source: &[char]) -> harper_core::Mask { |
| 29 | + let mut mask = Mask::new_blank(); |
| 30 | + |
| 31 | + let mut location = 0; |
| 32 | + let mut in_code_env = false; |
| 33 | + let mut last_line_blank = false; |
| 34 | + |
| 35 | + for line in source.split(|c| *c == '\n') { |
| 36 | + let string_form = line.to_string(); |
| 37 | + let trimmed = string_form.trim(); |
| 38 | + let line_is_bird = line.first().is_some_and(|c| *c == '>'); |
| 39 | + |
| 40 | + // Code fencing |
| 41 | + let latex_style = matches!(trimmed, r"\begin{code}" | r"\end{code}"); |
| 42 | + let code_start = trimmed == r"\begin{code}" || (last_line_blank && line_is_bird); |
| 43 | + let code_end = trimmed == r"\end{code}" || trimmed.is_empty(); |
| 44 | + |
| 45 | + // Toggle on fence |
| 46 | + if (!in_code_env && code_start) || (in_code_env && code_end) { |
| 47 | + in_code_env = !in_code_env; |
| 48 | + |
| 49 | + // Exclude latex-style fence |
| 50 | + if latex_style { |
| 51 | + location += line.len() + 1; // +1 for the newline split on |
| 52 | + last_line_blank = trimmed.is_empty(); |
| 53 | + continue; |
| 54 | + } |
| 55 | + |
| 56 | + // Exclude newline after code for bird style |
| 57 | + if trimmed.is_empty() { |
| 58 | + location += line.len() + 1; // +1 for the newline split on |
| 59 | + last_line_blank = true; |
| 60 | + continue; |
| 61 | + } |
| 62 | + } |
| 63 | + |
| 64 | + let end_loc = location + line.len(); |
| 65 | + if (!in_code_env && self.text) || (in_code_env && self.code) { |
| 66 | + let start_loc = if line_is_bird { location + 2 } else { location }; |
| 67 | + mask.push_allowed(Span::new(start_loc, end_loc)); |
| 68 | + } |
| 69 | + |
| 70 | + location = end_loc + 1; // +1 for the newline split on |
| 71 | + last_line_blank = trimmed.is_empty(); |
| 72 | + } |
| 73 | + |
| 74 | + mask.merge_whitespace_sep(source); |
| 75 | + mask |
| 76 | + } |
| 77 | +} |
| 78 | + |
| 79 | +#[cfg(test)] |
| 80 | +mod tests { |
| 81 | + use harper_core::{Masker, Span}; |
| 82 | + use itertools::Itertools; |
| 83 | + |
| 84 | + use super::LiterateHaskellMasker; |
| 85 | + |
| 86 | + #[test] |
| 87 | + fn bird_format() { |
| 88 | + let source = r"Text here |
| 89 | +
|
| 90 | +> fact :: Integer -> Integer |
| 91 | +> fact 0 = 1 |
| 92 | +> fact n = n * fact (n-1) |
| 93 | +
|
| 94 | +Text here |
| 95 | +" |
| 96 | + .chars() |
| 97 | + .collect_vec(); |
| 98 | + |
| 99 | + let text_mask = LiterateHaskellMasker::text_only().create_mask(&source); |
| 100 | + assert_eq!( |
| 101 | + text_mask |
| 102 | + .iter_allowed(&source) |
| 103 | + .map(|(s, _)| s) |
| 104 | + .collect_vec(), |
| 105 | + vec![Span::new(0, 10), Span::new(80, 90)], |
| 106 | + ); |
| 107 | + |
| 108 | + let code_mask = LiterateHaskellMasker::code_only().create_mask(&source); |
| 109 | + assert_eq!( |
| 110 | + code_mask |
| 111 | + .iter_allowed(&source) |
| 112 | + .map(|(s, _)| s) |
| 113 | + .collect_vec(), |
| 114 | + vec![Span::new(13, 39), Span::new(42, 52), Span::new(55, 78)], |
| 115 | + ); |
| 116 | + } |
| 117 | + |
| 118 | + #[test] |
| 119 | + fn latex_format() { |
| 120 | + let source = r#"Text here |
| 121 | +\begin{code} |
| 122 | +main :: IO () |
| 123 | +main = print "just an example" |
| 124 | +\end{code} |
| 125 | +Text here |
| 126 | +"# |
| 127 | + .chars() |
| 128 | + .collect_vec(); |
| 129 | + |
| 130 | + let text_mask = LiterateHaskellMasker::text_only().create_mask(&source); |
| 131 | + assert_eq!( |
| 132 | + text_mask |
| 133 | + .iter_allowed(&source) |
| 134 | + .map(|(s, _)| s) |
| 135 | + .collect_vec(), |
| 136 | + vec![Span::new(0, 9), Span::new(79, 89)], |
| 137 | + ); |
| 138 | + |
| 139 | + let code_mask = LiterateHaskellMasker::code_only().create_mask(&source); |
| 140 | + assert_eq!( |
| 141 | + code_mask |
| 142 | + .iter_allowed(&source) |
| 143 | + .map(|(s, _)| s) |
| 144 | + .collect_vec(), |
| 145 | + vec![Span::new(23, 67)], |
| 146 | + ); |
| 147 | + } |
| 148 | +} |
0 commit comments